Pypreprocess report for subj1

Start time: 19-Feb-2021 19:51:25

End time: 19-Feb-2021 19:53:06

Preprocessing steps:

All preprocessing was done using pypreprocess, a collection of python scripts and modules for preprocessing functional and anatomical MRI data.

For each subject, the following preprocessing steps have been done:

Motion correction has been done so as to estimate, and then correct for, subject's head motion.
The subject's functional images have been coregistered to their anatomical image. Coregistration is important as it allows: (1) segmentation of the functional via segmentation of the anatomical brain; (2) inter-subject registration via inter-anatomical registration, a trick referred to as 'Indirect Normalization'; (3) ROIs to be defined on the anatomy, making it possible for activation maps to be projected and appreciated thereupon.
Tissue Segmentation has been employed to segment the anatomical image into GM, WM, and CSF compartments, using template TPMs (Tissue Probability Maps).
The segmented anatomical image has been warped into the MNI template space by applying the deformations learnt during segmentation. The same deformations have been applied to the functional images.

References

Russel A. Poldrack et al. Handbook of Functional MRI Data Analysis

Visualisation for each step:

Motion Correction (see log)

Motion parameters estimated during motion-correction. If motion is less than half a voxel, it's generally OK. Moreover, it's recommended to include these estimated motion parameters as confounds (nuissance regressors) in the the GLM.

Corregistration mean_functional_image == > anatomical_image (see log)

The red contours should match the background image well. Otherwise, something might have gone wrong. Typically things that can go wrong include: lesions (missing brain tissue); bad orientation headers; non-brain tissue in anatomical image, etc. In rare cases, it might be that the registration algorithm simply didn't succeed.

Segmentation of _anatomical_image (see log)

Acronyms: TPM means Tissue Probability Map; GM means Grey-Matter; WM means White-Matter; CSF means Cerebro-Spinal Fuild. The TPM contours shoud match the background image well. Otherwise, something might have gone wrong. Typically things that can go wrong include: lesions (missing brain tissue); bad orientation headers; non-brain tissue in anatomical image (i.e needs brain extraction), etc. In rare cases, it might be that the segmentation algorithm simply didn't succeed.

Segmentation of _mean_functional_image (see log)

Segmentation of warped_anatomical_image (see log)

Segmentation of warped_mean_functional_image (see log)

Normalization of anatomical_image (see log)

Normalization of mean_functional_image (see log)

tsdiffana plot 0

(Squared) differences across sequential volumes. A large value indicates an artifact that occurred during the slice acquisition, possibly related to motion.

tsdiffana plot 1

Average signal over each volume. A large drop / peak (e.g. 1%) w.r.t the mean level indicates an artefact. For example, there are usually large values peaks in the first few slices due to T2 relaxation effects, and these slices are usually adviced to be discarded.

tsdiffana plot 2

Variance index per slice. Note that acquisition artifacts can be slice-specific. Look at the data if there is a peak somewhere.

tsdiffana plot 3

Scaled variance per slice indicates slices where artifacts occur. A slice/time with large variance should be eyeballed.

tsdiffana plot 4

Large variations should be confined to vascular structures or ventricles. Large variations around the brain indicate residual motion effects.

tsdiffana plot 5

Large variations should be confined to vascular structures or ventricles. Large variations around the brain indicate (uncorrected) motion effects.